Jeremy Green

Results 32 comments of Jeremy Green

It didn't check whether this was just caused by less frequent garbage collections in v8 than JavaScriptCore. I did file https://github.com/facebook/react-native/issues/27532 though, which might help in this situation. If close()...

@willholen - did you mean hermes in your above comment? (Hermes and v8 have the same symptoms, perhaps with the same cause, but this is an issue in the react-native-v8...

Hi, I wrote 52fbf0b, though otherwise have nothing to do with the puppet-jenkins project. I've just tried moving from using 52fbf0b to using release 1.7.0. With release 1.7.0 I found...

We have the same problem with another react-native library, so maybe the fix could be the same too: https://github.com/react-native-community/lottie-react-native/issues/443#issuecomment-478081871

We're being forced to use androidx by some other library that we use that dropped support for the support lib (androidx's predecessor).

In our app, we've seen a SEGV at libjsc.so:1013576 from at least Dec 28th 2019. It's affecting 0.2% of a large number of sessions, making it our most common crash....

I found the symbol by converting the offset to hex and looking it up in: ``` nm android/app/build/intermediates/merged_native_libs/productRelease/out/lib/arm64-v8a/libjsc.so | sort | ${ANDROID_HOME}/ndk-bundle/toolchains/llvm/prebuilt/darwin-x86_64/bin/arm-linux-androideabi-c++filt | less ```

Decoding some more of the backtrace: 1013576 JSC::AccessCase::propagateTransitions(JSC::SlotVisitor&) const 1327084 JSC::PolymorphicAccess::propagateTransitions(JSC::SlotVisitor&) const . . 1982884 WTF::SharedTaskFunctor::run() . . 6581708 WTF::wtfThreadEntryPoint(void*) ...it's still matching up with what's in this issue's description.

This seems to propose a viable fix: https://bugs.webkit.org/show_bug.cgi?id=200983 This mentions the problem too: https://bugs.webkit.org/show_bug.cgi?id=202150

@Kudo and @kmagiera, Do you know if there's any chance of react native accepting a new JavaScriptCore build? If so, would it be better to patch the existing WebKit version...